Robotics is a components business first

Every few months a humanoid walks across a stage, and the question that follows is which platform will win. We think that is the wrong question to ask this decade, or at least the wrong one to own.

Look at the bill of materials instead. A capable arm is harmonic drives and planetary reducers, precision bearings, ball screws, rare-earth magnets, force and torque sensing, and the power electronics to move all of it without cooking itself. Those parts are made by a small number of firms with long qualification cycles, and they are sold into every platform regardless of which one eventually ships in volume.

That asymmetry is the point. The platform layer is a winner-take-most contest with many entrants and unclear economics. The component layer is a supplier business with pricing power, incumbency and a customer list that grows with the category rather than with any single participant. One requires you to be right about the winner. The other only requires the category to be real.

Reliability, not capability, is what gates deployment. A machine that works in a demonstration and fails at ten thousand cycles is a research result, not a product, and closing that gap is a materials and mechanical problem before it is a software one. It is also the part of the problem that takes the longest and rewards the incumbents.

We hold the layer that is paid either way, and we are patient about the rest. This note is informational only, and not investment advice.